Odds on the Greenhouse

AS FAR AS HISTORIANS can tell, the first attempt to predict the weather through mathematical equations came in 1910, when English physicist Lewis F. Richardson divided Britain and part of Western Europe into a grid with squares 120 miles on a side. He collected observations--humidity, temperature, winds--at specified altitudes in each square, then calculated (by hand) the interactions that would create the next day's weather. His forecast: superhurricane winds in excess of 100 mph over England on May 20.

They didn't blow. So much for mathematical modeling.

No wonder that the scientists who most doubt the threat of climate change due to the greenhouse effect are meteorologists. If they can't get it right 72 or even 24 hours ahead, how can anyone forecast climate 100 years ahead? Coal, oil and auto companies have long played up the unreliability of climate forecasts, which say that once the atmosphere contains twice as much carbon dioxide as it did in 1870, the world will be 1.8 to 8.1 degrees warmer and suffer punishing bouts of drought, floods and sea-level rise. But a week before some 160 countries meet in Kyoto, Japan, to negotiate controls on the gases that cause this greenhouse effect, industry is turning up the volume. ""We are swimming in a sea of uncertainty,'' said William O'Keefe of the American Petroleum Institute at a forum in New York last week. Now a respected climatologist has waded into that sea to quantify the uncertainties. ""[Some climate] projections have a greater than 99 out of 100 chance of being true,'' concludes Jerry Mahlman, director of the Geophysical Fluid Dynamics Laboratory in Princeton, N.J., in the current issue of the journal Science. ""Some are just wrong.''

In his ""just wrong'' category is the prediction that the number of tropical storms and hurricanes will increase. This worry has spurred some in the insurance industry--memories of 1992's Hurricane Andrew are still painful--to push for greenhouse-gas controls. Nor is there ""credible scientific support,'' Mahlman argues, for the idea that thunderstorms outside the tropics will intensify. More certain, he says, is the forecast that in a warmer world soil moisture in the American grain belt and other northern, midlatitude regions will decrease, because of hotter summers. (Higher temps cause more evaporation; more evaporation dries the soil.) But because soil moisture also depends on snowpack and on spring and summer rains, which are notoriously hard to predict, this forecast has only a two- in-three chance of coming true. Mahlman gives the same odds that tropical storms will grow more intense than they do today.

The core forecast of the climate-change models--global warming of 2.7 to 8.1 degrees by 2060 or so--is a good bet: Mahlman gives it a 9 in 10 chance of being right. The big temperature spread partly reflects uncertainty (in the computer models used to forecast climate) of how a warming world would affect clouds. More evaporation should mean more clouds; high clouds cool the planet. But cloud formation is poorly understood. If more form, or if they cool the earth more than scientists think, greenhouse warming would be at the low end of the prediction; if fewer form, or form lower, warming would be more extreme. Also uncertain: what aerosol pollution, also from coal and oil, will do to temperatures.

Mahlman's attempt to play climatological bookie is based on Atmospheric General Circulation Models. AGCMs are computer models consisting of 40,000 to 120,000 lines of code for the fundamental laws of physics. They include equations ranging from Newton's laws of motion (how and whether winds pick up) to classical gas laws (expanding a mass of air cools it, and warmer air holds more water vapor). ""Climate models are simpler than people think,'' says Mahlman, ""even though they have an outrageous number of computations.''

And they've improved. Even a decade ago, says Piers Sellers, a biometeorologist at NASA's Johnson Space Center, various models gave wildly divergent results. Now ""the results are converging.'' One improvement has been to incorporate aerosols. They tend to cool things off. Without them, the models predicted roughly 2 degrees of warming over the last century rather than the 1 degree that actually occurred. More precision came from including how much solar radiation deserts and forests reflect. In the latest refinement, AGCMs include equations describing how plants suck up carbon dioxide and radiation, change the movement of air masses and evaporate water (water vapor is a greenhouse gas, so more evaporation brings more warming). Yet there is a sad irony in the increased precision of the climate models. On what to do about greenhouse gases, a decision that turns on values and justice, the mathematical models offer no answers.